Message-Id: <v02130503ae937d71cb0a@[128.120.185.208]>
In-Reply-To: <199610221429.KAA11995@sarah.albany.edu>
Date: Wed, 23 Oct 1996 02:23:08 -0700
To: linconf@tamvm1.tamu.edu
From: pmfarrell@ucdavis.edu (Patrick Farrell)
Subject: Reply to Broadwell from Farrell
FARRELL REPLY TO BROADWELL
Broadwell suggests that my comments about the argumentation in his paper
essentially amounted to pointing out that it rests largely on certain
assumptions of the principles and parameters approach. This is correct with
respect to my second point, which was, essentially, that although the data
presented seemed to indicate some kind of subject orientation for
switch-reference marking, they did not directly indicate the need for an
account in terms of "geometric configuration". As for the first point,
however, which was that the "raised" possessor in the so-called possessor
raising construction is simply assumed to not be an argument of the verb of
which it is subject, I don't think the Projection Principle or any other
aspect of the principles and parameters approach, per se, yields this
assumption. Consider the following sentences:
a. I touched my sister on the arm.
b. I touched my sister's arm.
It seems to me that these are truth-conditionally equivalent and differ
only in the topicality or prominence of the NP _my sister_. But in (a)
_touch_ has two syntactic complements, whereas it only has one in (b).
Since the Projection Principle does not allow changing the
subcategorization of a verb by a syntactic process, (a) cannot be derived
from (b) by possessor raising. Hence, it must be possible for _touch_ to
have two different subcategorization frames and hence two different
argument arrays. Now, given the possibility of this kind of argument
augmentation (even assuming standard Principles and Parameters stuff, I
think), it is an empirical question whether the same kind of thing occurs
or not in the Choctaw and Chickasaw cases that Broadwell discusses. If I
were a speaker of a language that didn't allow alternations like that in
(a)-(b), it seems that I might give one gloss for the two sentences. I
can't help but wonder if it is simply the limitations of English that make
it reasonable to provide one gloss for sentences like:
c.) [Jan im-aaimpa' iyy]-at oppolo.
Jan III-table leg-NM broken
d.) [Jan im-aaimp]-at [iyy]-at oppolo.
Jan III-table-NM leg-NM broken
e.) [Jan]-at [im-aaimp]-at [iyy]-at oppolo.
Jan-NM III-table-NM leg-NM broken.
'The leg of Jan's table is broken.'
Can we say "Clearly _Jan_ and _aaimp_ are not arguments of _oppolo_ in
(e)"? (a) seems to differ from (b) in that it somehow highlights the effect
of the touching act on my sister (which seems to be another way of saying
that there is some kind of difference in topicality). If the leg of Jan's
table is broken, the breaking event affects the leg, the table, and Jan.
Could it not be that (e) differs from (c) in that it somehow highlights the
effect of the breaking event on Jan and the table?
Without a clear and explicit definition of "argument" and criteria for
determining whether a given constituent is one in Choctaw and Chickasaw, I
don't see how it is possible to conclude confidently from the existence of
sentences such as (c)-(e) (and their switch-reference properties) that
argumenthood cannot possibly be relevant to switch-reference marking, even
given standard principles and parameters assumptions.
***********
Patrick Farrell
Linguistics Program
UC Davis